Getting Started with MCP Catie

Introduction to MCP Catie

MCP Catie (Context Aware Traffic Ingress Engine) is a lightweight, configurable proxy service designed specifically for routing MCP (Model Context Protocol) JSON-RPC requests to appropriate backend services based on request content.

What is MCP?

The Model Context Protocol (MCP) is a JSON-RPC based protocol for communication between clients and AI models. It provides a standardized way to:

Request completions from language models
Access tools and resources
Manage conversation context
Stream responses in real-time

Why Use MCP Catie?

MCP Catie solves several challenges when working with MCP services:

Dynamic Routing: Route requests to different backends based on content
Session Management: Maintain client connections to the same backend
Streamable Transport: Support for SSE (Server-Sent Events)
Observability: Built-in monitoring and metrics

Key Features

Content-Based Routing: Route requests based on method and parameters
Pattern Matching: Use regular expressions to match resource URIs and tool names
Session Awareness: Track client sessions to maintain connection consistency
Hot Configuration: Reload configuration without service restart
Monitoring: Built-in UI and Prometheus metrics
Containerization: Easy deployment with Docker

Current Status

MCP Catie is currently in alpha status. While it's functional for basic use cases, you may encounter issues or missing features. We welcome feedback and contributions to improve the project.

In the next sections, we'll cover installation, configuration, and usage of MCP Catie.